Network element fault detection method and device, computer equipment and storage medium

Abstract

The invention relates to a network element fault detection method and device, computer equipment and a storage medium, and the method comprises the steps: predicting a future data sequence of a target network element in a future time period according to a current data sequence of the target network element in a current time period; determining a current global error between the current data sequence and the future data sequence; processing the historical data sequence of the target network element in the first historical time period to obtain a basic threshold value; determining a first correction value according to the historical data sequence of the target network element in the global historical period and the network element evaluation index data at the current moment and the historical data sequence of the associated network element of the target network element in the global historical period and the network element evaluation index data at the current moment; adjusting the basic threshold value by adopting the first correction value to obtain a dynamic threshold value; and determining a fault detection result of the target network element according to a size relationship between the current global error and the dynamic threshold. The method can adapt to a dynamically changing communication network environment, and the accuracy of network element fault detection is improved.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a network element fault detection method, apparatus, computer equipment, and storage medium. Background Art

[0002] In communications networks, network elements (NEs) are the basic functional units that make up the network, such as base stations, switches, routers, optical transmission NEs, core network NEs, firewalls, and servers. NEs are responsible for the transmission, routing, switching, and processing of data within the communications network. Their design and deployment directly impact the performance, reliability, and scalability of the network.

[0003] With the rapid development of technologies such as 5G (fifth-generation mobile communication technology), the Internet of Things (IoT), and cloud computing, network traffic on communications networks has exploded, and the types of services on these networks are becoming increasingly diverse. This has led to significant fluctuations in the service indicators faced by network elements (NEs). To ensure the normal operation of communication networks, unified management of NEs is required to promptly detect NE failures and maintain communication security.

[0004] In traditional technology, generative AI (Artificial Intelligence) is used to generate network element evaluation indicators and determine anomalies for network element fault detection. The determination method is based on the error between the predicted data and the actual data of the network element evaluation indicators.

[0005] However, in the above-mentioned traditional technology, a static threshold value of a network element evaluation index is used as a judgment criterion, which is difficult to adapt to the dynamically changing communication network environment, resulting in low accuracy of network element fault detection. Summary of the Invention

[0006] Based on this, it is necessary to provide a network element fault detection method, device, computer equipment and storage medium to address the above technical problems, which can adapt to the dynamically changing communication network environment and improve the accuracy of network element fault detection.

[0007] In a first aspect, the present application provides a network element fault detection method, comprising:

[0008] Predicting a future data sequence of the target network element in a future time period based on the current data sequence of the target network element in the current time period; wherein the data sequence includes network element evaluation index data of at least two consecutive moments;

[0009] Determine the current global error between the current data sequence and the future data sequence;

[0010] Processing a historical data sequence of the target network element in a first historical period to obtain a basic threshold value; wherein the first historical period is a period when the target network element is in a normal state;

[0011] Determining a first correction value based on a historical data sequence of the target network element in a global historical period and network element evaluation index data at a current moment, and a historical data sequence of network elements associated with the target network element in a global historical period and network element evaluation index data at a current moment; wherein the global historical period is a historical period corresponding to the current period, and the global historical period includes the first historical period;

[0012] The basic threshold is adjusted using the first correction value to obtain a dynamic threshold;

[0013] The fault detection result of the target network element is determined based on the size relationship between the current global error and the dynamic threshold.

[0014] In one embodiment, processing a historical data sequence of a target network element in a first historical period to obtain a basic threshold value includes:

[0015] Based on the data generation model, predicting the predicted data sequence of the target network element in the second historical period according to the historical data sequence of the target network element in the first historical period;

[0016] Constructing a historical error sequence based on the actual data sequence and the predicted data sequence of the target network element in the second historical period;

[0017] Performing statistical analysis on the historical error sequence to obtain the historical local characteristics of the historical error sequence; wherein the historical local characteristics include the historical local error mean and the historical local error standard deviation;

[0018] Determine the basic threshold based on historical local characteristics.

[0019] In one embodiment, determining the first correction value based on a historical data sequence of a target network element in a global historical period and network element evaluation index data at a current moment, and a historical data sequence of network elements associated with the target network element in a global historical period and network element evaluation index data at a current moment, includes:

[0020] Determine the error weight value of the target network element based on the historical data sequence of the target network element in the global historical period and the network element evaluation index data at the current moment;

[0021] Determine the error weight value of the associated network element based on the historical data sequence of the associated network element of the target network element in the global historical period and the network element evaluation index data at the current moment;

[0022] A first correction value is determined according to a mean value between the weighted error value of the target network element and the weighted error value of the associated network element.

[0023] In one embodiment, determining an error weighted value of a target network element based on a historical data sequence of the target network element in a global historical period and network element evaluation index data at a current moment includes:

[0024] Performing statistical analysis on the historical data series of the target network element in the global historical period to obtain the historical global characteristics of the target network element; wherein the historical global characteristics include the historical global error mean and the historical global error standard deviation;

[0025] Determine the real-time error of the target network element at the current moment according to the difference between the network element evaluation index data of the target network element at the current moment and the network element evaluation index data at the future moment corresponding to the current moment;

[0026] The error weighted value of the target network element is determined according to the historical global characteristics, real-time error and network element weight of the target network element.

[0027] In one embodiment, determining an error weighted value of an associated network element according to a historical data sequence of an associated network element of the target network element in a global historical period and network element evaluation index data at a current moment includes:

[0028] Performing statistical analysis on the historical data series of the associated network elements in the global historical period to obtain historical global characteristics of the associated network elements; wherein the historical global characteristics include the historical global error mean and the historical global error standard deviation;

[0029] Determine the real-time error of the associated network element at the current moment according to the difference between the network element evaluation index data of the associated network element at the current moment and the network element evaluation index data at the future moment corresponding to the current moment;

[0030] The error weighted value of the associated network element is determined according to the historical global characteristics, real-time error and network element weight of the associated network element.

[0031] In one embodiment, adjusting the basic threshold using the first correction value to obtain the dynamic threshold includes:

[0032] Determining a current error standard deviation based on the first sub-data sequence and a second sub-data sequence in the future data sequence that corresponds to the first sub-data sequence; wherein the first sub-data sequence is a data sequence in the current data sequence that includes network element evaluation indicator data at the current moment;

[0033] Determining a second correction value based on the current error standard deviation and the historical local error standard deviation;

[0034] Determining a third correction value based on whether the current time period is within a designated event period and a designated event coefficient; wherein the designated event coefficient is determined based on a historical data sequence of the target network element within the designated event period; the designated event includes a network element cutover event and / or a non-working day event;

[0035] The basic threshold is adjusted using the first correction value and the auxiliary value to obtain a dynamic threshold; wherein the auxiliary value includes the second correction value and / or the third correction value.

[0036] In a second aspect, the present application further provides a network element fault detection device, comprising:

[0037] A data prediction module is used to predict the future data sequence of the target network element in the future time period based on the current data sequence of the target network element in the current time period; wherein the data sequence includes network element evaluation index data of at least two consecutive moments;

[0038] an error determination module for determining the current global error between the current data sequence and the future data sequence;

[0039] a threshold determination module, configured to process a historical data sequence of the target network element in a first historical period to obtain a basic threshold; wherein the first historical period is a period when the target network element is in a normal state;

[0040] a correction value determination module, configured to determine a first correction value based on a historical data sequence of the target network element in a global historical period and network element evaluation index data at a current moment, and a historical data sequence of network elements associated with the target network element in a global historical period and network element evaluation index data at a current moment; wherein the global historical period is a historical period corresponding to the current period, and the global historical period includes the first historical period;

[0041] A threshold adjustment module, configured to adjust the basic threshold using the first correction value to obtain a dynamic threshold;

[0042] The fault detection module is used to determine the fault detection result of the target network element according to the size relationship between the current global error and the dynamic threshold.

[0043] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps in the various method embodiments provided in the first aspect when executing the computer program.

[0044] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the various method embodiments provided in the first aspect above.

[0045] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps in the various method embodiments provided in the first aspect above.

[0046] The above-mentioned network element fault detection method, device, computer equipment and storage medium predict the future data sequence of the target network element in the future time period based on the current data sequence of the target network element in the current time period, wherein the above-mentioned data sequence includes network element evaluation index data at at least two moments, so that the current global error between the above-mentioned current data sequence and the future data sequence can be determined; then, the historical data sequence of the first historical time period when the target network element is in a normal state is processed to obtain a basic threshold, and the first correction value is determined based on the historical data sequence of the global historical time period corresponding to the current time period and including the first historical time period and the network element evaluation index data at the current moment, as well as the historical data sequence of the associated network elements of the target network element in the global historical time period and the network element evaluation index data at the current moment; and then the first correction value is used to adjust the basic threshold to obtain a dynamic threshold, and the fault detection result of the target network element is determined based on the size relationship between the current global error and the dynamic threshold. In this way, on the one hand, the historical data sequence of the first historical period when the target network element was in a normal state is used to obtain the basic threshold, providing a basic level for network element fault detection. On the other hand, because the global historical period corresponds to the current period, the global historical period changes with the current period. Therefore, the historical data sequence of the target network element's associated network elements in the global historical period is used to determine the first correction value used to adjust the basic threshold. The basic threshold is adjusted using the first correction value. The resulting threshold for network element fault detection is a dynamic threshold, which improves the threshold sensitivity and avoids false alarms caused by the isolation of the target network element. Based on this, the use of a dynamic threshold determined by integrating historical data sequences and multi-network element association can adapt to dynamically changing communication network environments, provide more comprehensive and accurate detection of network element faults, and improve the accuracy of network element fault detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0048] Figure 1 A diagram illustrating an application environment of a network element fault detection method provided in some embodiments of the present application;

[0049] Figure 2A flowchart of a network element fault detection method provided in some embodiments of the present application;

[0050] Figure 3 A schematic diagram of a process for determining a basic threshold value provided in some embodiments of the present application;

[0051] Figure 4 A schematic diagram of a process for determining a first correction value provided in some embodiments of the present application;

[0052] Figure 5 A schematic diagram of a process for determining an error weighted value of a target network element provided in some embodiments of the present application;

[0053] Figure 6 A schematic diagram of a process for determining an error weighted value of an associated network element provided in some embodiments of the present application;

[0054] Figure 7 A schematic diagram of a flow chart for determining a dynamic threshold value provided in some embodiments of the present application;

[0055] Figure 8 A flowchart of a network element fault detection method provided in some other embodiments of the present application;

[0056] Figure 9 A structural block diagram of a network element fault detection device provided in some embodiments of the present application;

[0057] Figure 10 An internal structural diagram of a computer device provided for some embodiments of the present application. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0059] It should be noted that the terms "first", "second", etc. used in this application may be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "including" and "having" used in this application and any variations thereof are intended to cover non-exclusive inclusions. The term "plurality" used in this application refers to two or more. The term "and / or" used in this application refers to one or any combination of multiple solutions.

[0060] Traditional technologies use generative AI (artificial intelligence) to generate network element evaluation indicators and identify anomalies for network element fault detection. This determination is based on the difference between the predicted and actual data of the network element evaluation indicators. However, these traditional technologies use static thresholds for network element evaluation indicators as the criteria for determination, making them difficult to adapt to dynamically changing communication network environments, resulting in low accuracy in network element fault detection.

[0061] To address the above technical issues, in an exemplary embodiment, a network element fault detection method is provided. The method can be applied to a server. The server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services, such as a network management system developed by a manufacturer or operator.

[0062] In an exemplary embodiment, a network element fault detection method provided in an embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the network element 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. Among them, the data storage system can store real network element evaluation index data and predicted network element evaluation index data of each network element in the communication network. In this way, the server 104 can obtain the current data sequence of the target network element in the current time period, the historical data sequence of the target network element in the global historical time period, and the historical data sequence of the associated network elements of the target network element in the global historical time period from the data storage system to execute the network element fault detection method provided in the embodiment of the present application to perform fault detection on the target network element.

[0063] In an exemplary embodiment, Figure 2 As shown, a network element fault detection method is provided, which is applied to Figure 1 Taking the server 104 in the example as an example, the following steps may be included:

[0064] S201 , predicting a future data sequence of a target network element in a future time period based on a current data sequence of the target network element in a current time period.

[0065] The data sequence includes network element evaluation index data at at least two consecutive moments.

[0066] Network element evaluation metrics are core parameters used to measure the operational status, performance, reliability, and resource utilization of network elements. These metrics typically include CPU (Central Processing Unit) utilization, memory usage, storage usage, port / bandwidth utilization, concurrent connections, signaling success rate, data validation accuracy, routing convergence time, synchronization precision, data processing success rate, packet loss rate, forwarding rate, bit error rate (BER), latency, jitter, and retransmission rate.

[0067] The so-called current period is a time range with the current moment as the end moment and a preset duration. Correspondingly, the so-called future period is a time range with the moment after the current moment as the start moment and a preset duration. That is, the end moment of the current period and the start moment of the future period are consecutive moments, the current period and the future period have the same time length, and the current data sequence and the future data sequence include the same amount of network element evaluation indicator data. The above-mentioned preset duration can be limited based on empirical values, test values ​​from multiple tests, and actual application requirements, and is not specifically limited in this regard.

[0068] The so-called network element evaluation indicator data may be the indicator value of the network element evaluation indicator, such as the specific indicator value of CPU utilization, packet loss rate, and forwarding rate, or may be data used to evaluate the network element operation status determined based on the indicator value of the network element evaluation indicator, such as data obtained by normalizing the specific indicator values ​​of CPU utilization, packet loss rate, and forwarding rate and then performing a weighted summation. This is not specifically limited.

[0069] As can be seen from the above, when fault detection is performed on the target network element, the index value of the network element evaluation index of the target network element at at least two consecutive moments in the current time period is first obtained. For example, the target network element can have a built-in SNMP (Simple Network Management Protocol) agent to actively or passively report the index value of the network element evaluation index of the target network element to the sending server at the request of the server. For another example, the network management system as the execution subject directly collects the index value of the network element evaluation index from the network element; wherein, the network element type of the target network element can be set according to the needs of the actual application, and then the type of the network element evaluation index can be set, and the network element evaluation index can be one or more, and there is no specific limitation on this. After obtaining the network element evaluation index data, the network element evaluation index data of the target network element at at least two consecutive moments in the current time period is determined based on the obtained index value. For example, the index value of the network element evaluation index at each moment is used as the network element evaluation index data at that moment. For another example, the index values ​​of multiple network element evaluation indicators at each moment are normalized and weighted summed, and the obtained sum value is used as the network element evaluation index data at that moment.

[0070] Optionally, the index values ​​of the network element evaluation index of the target network element at at least two consecutive moments in the current time period are obtained, and the obtained index values ​​are initialized to obtain initialized index values. The initialization processing at least includes data cleaning, formatting conversion, and normalization processing.

[0071] Among them, the original indicator values ​​obtained may contain outliers, missing values, duplicate values ​​and other problems, and it is necessary to ensure the accuracy and consistency of the data through cleaning, for example, filtering outliers, filling in missing values, merging duplicate values, etc.

[0072] The indicator values ​​of different gateway evaluation indicators may be obtained through multiple protocols. Therefore, the formats of the indicator values ​​of different gateway evaluation indicators vary greatly and need to be converted into a unified format for easy analysis, such as data type conversion, structure standardization, and multi-source data fusion.

[0073] The physical meanings and numerical ranges of the indicator values ​​of different gateway evaluation indicators vary greatly (such as CPU utilization ∈ [0, 100%], forwarding delay ∈ [0, 1000ms], etc.), and need to be mapped to a unified interval (such as [0, 1] or [-1, 1]) through normalization to facilitate comprehensive analysis (such as determining network element evaluation indicator data), for example, linear normalization, standardization, logarithmic normalization, reverse indicator processing, etc.

[0074] After obtaining the current data sequence of the target network element in the current time period, the future data sequence of the target network element in the future time period can be predicted based on the current data sequence of the target network element in the current time period.

[0075] Optionally, a sample network element is selected and a historical period in which the sample network element was in a normal state is determined as a sample historical period. Network element evaluation index data for the sample network element at at least two consecutive moments in the sample historical period is obtained as a sample current data sequence, and network element evaluation index data for the sample network element at at least two consecutive moments in a future period of the sample historical period is obtained as a sample future data sequence. The sample current data sequence and the sample future data sequence contain the same number of network element evaluation index data, and the temporal relationship between the sample historical period and the future period of the sample historical period is the same as the temporal relationship between the current period and the future period. In this way, a data generation model can be trained using the sample current data sequence of the sample network element as input and the sample future data sequence of the sample network element as a label. After obtaining the current data sequence of the target network element in the current period, the current data sequence is input into the data generation model to obtain, as output by the data generation model, a future data sequence for the target network element in the future period. The duration of the sample historical periods of different sample network elements can be the same or different. Optionally, the duration of the sample historical period of at least one sample network element is the same as the duration of the current period.

[0076] S202, determining a current global error between a current data sequence and a future data sequence.

[0077] As mentioned above, the current data sequence and the future data sequence include the same number of network element evaluation index data. Therefore, the current data sequence and the future data sequence are equivalent to two one-dimensional series of the same length. The error between the current data sequence and the future data sequence can be determined to obtain the current global error between the current data sequence and the future data sequence.

[0078] The current global error is a core indicator that measures the degree of difference between the current data sequence and the future data sequence and is used to perform deviation analysis on the current data sequence and the future data sequence. Furthermore, the current global error can be any error indicator that can measure the degree of difference between the current data sequence and the future data sequence, such as Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Relative Error (MRE), Symmetric Mean Absolute Percentage Error (SMAPE), Sum of Squared Errors (SSE), Mean Relative Error (MRE), Symmetric Mean Absolute Percentage Error (SMAPE), Sum of Squared Errors (SSE), Correlation Coefficient, etc.

[0079] Optionally, a current error sequence is constructed based on the current data sequence and the future data sequence; the current error sequence is statistically analyzed to obtain the current local features of the current error sequence as the current global error between the current data sequence and the future data sequence; wherein, the element value of the i-th element in the above-mentioned current error sequence is the difference between the element value of the i-th element in the current data sequence and the element value of the i-th element in the future data sequence, and the above-mentioned current local features can be any error indicator of a "global feature" that can reflect the overall trend, amplitude or fluctuation of each element in the current error sequence, such as mean error, mean absolute error, mean square error, root mean square error and standard deviation.

[0080] S203: Process the historical data sequence of the target network element in the first historical period to obtain a basic threshold.

[0081] The first historical period is a period when the target network element is in a normal state.

[0082] The so-called normal state of a network element means that all operating indicators of the network element are within the preset normal threshold range, it can perform its own functions stably and efficiently, and no abnormal phenomena that affect service carrying or network performance occur. Therefore, for the target network element, a time period before the current moment and in which the target network element is in a normal state can be determined as the first historical period, and a historical data sequence of the target network element in the first historical period can be obtained, wherein the historical data sequence of the target network element in the first historical period includes network element evaluation indicator data of at least two consecutive moments, and the acquisition method is the same as the acquisition method of the above-mentioned current data sequence.

[0083] The historical data sequence of the target network element in the first historical period can then be processed to obtain a basic threshold. Optionally, the mean, mean absolute error, mean square error, root mean square error, or standard deviation of each network element evaluation indicator data in the historical data sequence of the target network element in the first historical period can be determined as the basic threshold.

[0084] The duration of the first historical period can be limited based on experience, test values ​​from multiple tests, and actual application requirements, and no specific limitation is imposed on this.

[0085] S204, determining a first correction value based on the historical data sequence of the target network element in the global historical period and the network element evaluation index data at the current moment, and the historical data sequence of the network elements associated with the target network element in the global historical period and the network element evaluation index data at the current moment.

[0086] The global historical period is the historical period corresponding to the current period, and the global historical period includes the first historical period.

[0087] The so-called global historical period ends at the time before the start time of the current period, has a specified duration, and includes the time range of the first historical period. The specified duration can be limited based on empirical values, test values ​​from multiple experiments, and actual application requirements, and the specified duration must not be less than the duration of the first historical period. That is, the first historical period is a sub-period of the global historical period, and the global historical period includes a sub-period in which at least one target network element is in a normal state.

[0088] Optionally, the end time of the global historical period can be the moment before the start time of the current period, so that the end time of the global historical period and the start time of the current period are consecutive moments. Alternatively, the end time of the global historical period is before the start time of the current period and is separated from the start time of the current period by a target duration. The above target duration can be defined based on empirical values, experimental values ​​from multiple tests, and actual application requirements.

[0089] Optionally, after determining the current period, first determine the global historical period corresponding to the current period, and then select a sub-period in which the target network element is in a normal state from the global historical period as the first historical period.

[0090] Associated NEs of a target NE are other NEs that are directly or indirectly associated with the target NE in terms of network architecture, service processes, data interaction, or resource dependencies. These associations may be physical connections (e.g., link layer connections), logical interactions (e.g., protocol communication), or service collaboration (e.g., data forwarding). The core purpose is to influence the operating status of the target NE or be influenced by it through these associations.

[0091] Optionally, the target network element may be associated with one or more network elements.

[0092] The so-called current moment is the last moment of the current period. Correspondingly, the network element evaluation index data at the current moment is the last network element evaluation index data in the current data sequence.

[0093] Thus, after determining the above-mentioned global historical period, the historical data sequence of the target network element in the global historical period and the historical data sequence of the target network element's associated network elements in the global historical period can be obtained. The above-mentioned historical data sequence of the target network element in the global historical period and the historical data sequence of the target network element's associated network elements in the global historical period include network element evaluation indicator data at at least two consecutive moments, and the acquisition method is the same as the acquisition method of the above-mentioned current data sequence.

[0094] Then, the network element evaluation index data of the associated network elements of the target network element at the current moment is obtained; in this way, the first correction value can be determined based on the historical data sequence of the target network element in the global historical period and the network element evaluation index data at the current moment, as well as the historical data sequence of the associated network elements of the target network element in the global historical period and the network element evaluation index data at the current moment.

[0095] Optionally, determine the mean and standard deviation of the evaluation index data of each network element in the historical data sequence of the target network element in the global historical period to obtain a first mean and a first standard deviation, and determine the first correction parameter value based on the first mean and the second standard deviation, as well as the network element evaluation index data of the target network element at the current moment; determine the mean and standard deviation of the evaluation index data of each network element of the associated network elements of the target network element in the historical data sequence of the global historical period to obtain a second mean and a second standard deviation, and determine the second correction parameter value based on the second mean and the second standard deviation, as well as the network element evaluation index data of the associated network elements of the target network element at the current moment; and then determine the mean of the first correction parameter value and the second correction parameter as the first correction value.

[0096] S205: Use the first correction value to adjust the basic threshold to obtain a dynamic threshold.

[0097] After obtaining the first correction value and the basic threshold, the first correction value may be used to adjust the basic threshold to obtain the dynamic threshold.

[0098] Alternatively, the sum of the first correction value and the basic threshold value may be determined as the dynamic threshold value, or the difference between the basic threshold value and the first correction value may be determined as the dynamic threshold value.

[0099] S206 , determining a fault detection result of the target network element according to a magnitude relationship between the current global error and the dynamic threshold.

[0100] After obtaining the current global error and the dynamic threshold, the magnitude relationship between the current global error and the dynamic threshold can be determined, and the fault detection result of the target network element can be determined based on the magnitude relationship.

[0101] Optionally, when the current global error is greater than the dynamic threshold, the fault detection result of the target network element is that the target network element has a fault, that is, the target network element has a fault during the current time period. Correspondingly, when the current global error is less than or equal to the dynamic threshold, the fault detection result of the target network element is that the target network element has no fault, that is, the target network element is in a normal state during the current time period.

[0102] In the above-mentioned network element fault detection method, the future data sequence of the target network element in the future time period is first predicted based on the current data sequence of the target network element in the current time period, wherein the above-mentioned data sequence includes network element evaluation index data at at least two moments, so that the current global error between the above-mentioned current data sequence and the future data sequence can be determined; then, the historical data sequence of the first historical time period when the target network element is in a normal state is processed to obtain a basic threshold, and the first correction value is determined based on the historical data sequence of the global historical time period corresponding to the current time period and including the first historical time period and the network element evaluation index data at the current moment, as well as the historical data sequence of the associated network elements of the target network element in the global historical time period and the network element evaluation index data at the current moment; and then the first correction value is used to adjust the basic threshold to obtain a dynamic threshold, and the fault detection result of the target network element is determined based on the size relationship between the current global error and the dynamic threshold. In this way, on the one hand, the historical data sequence of the first historical period when the target network element was in a normal state is used to obtain the basic threshold, providing a basic level for network element fault detection. On the other hand, because the global historical period corresponds to the current period, the global historical period changes with the current period. Therefore, the historical data sequence of the target network element's associated network elements in the global historical period is used to determine the first correction value used to adjust the basic threshold. The basic threshold is adjusted using the first correction value. The resulting threshold for network element fault detection is a dynamic threshold, which improves the threshold sensitivity and avoids false alarms caused by the isolation of the target network element. Based on this, the use of a dynamic threshold determined by integrating historical data sequences and multi-network element association can adapt to dynamically changing communication network environments, provide more comprehensive and accurate detection of network element faults, and improve the accuracy of network element fault detection.

[0103] On the basis of the above embodiments, in an exemplary embodiment, the determination of the above basic threshold is further refined, optionally, as follows: Figure 3 As shown, the following steps may be included:

[0104] S301 , based on a data generation model, predicting a predicted data sequence of a target network element in a second historical period according to a historical data sequence of the target network element in a first historical period.

[0105] In this embodiment, as described above, a sample network element can be selected and a sample historical period and a future historical period of the sample historical period can be determined for the sample network element. This allows obtaining a sample current data sequence of the sample network element in the sample historical period and a sample future data sequence of the sample network element in a future period of the sample historical period. Furthermore, a data generation model can be trained using the sample current data sequence of the sample network element as input and the sample future data sequence of the sample network element as a label.

[0106] Thus, after obtaining the historical data sequence of the target network element in the first historical period, the historical data sequence of the target network element in the first historical period can be input into the above-mentioned data generation model to obtain the predicted data sequence of the target network element in the second historical period output by the above-mentioned data generation model. The historical data sequence of the target network element in the first historical period and the predicted data sequence of the target network element in the second historical period include the same number of network element evaluation indicator data.

[0107] Since the sample historical period of the sample network element and the future historical period of the sample historical period have the same duration, the first historical period and the second historical period have the same duration, and the second historical period is the period after the first historical period. Therefore, the time relationship between the sample historical period and the future period of the sample historical period, the time relationship between the first historical period and the second historical period, and the time relationship between the current period and the future period are all the same.

[0108] Optionally, the above-mentioned global historical period may include the above-mentioned second historical period.

[0109] Optionally, the first historical period is the same length as the current period.

[0110] S302: Construct a historical error sequence based on the actual data sequence and the predicted data sequence of the target network element in the second historical period.

[0111] It will be appreciated that the predicted data sequence for the target network element in the second historical period is based on the target network element's historical data sequence in the first historical period, rather than the target network element's actual data sequence in the second historical period. However, the second historical period is the historical period before the current moment, so an actual data sequence for the target network element in the second historical period exists. Thus, a historical error sequence is constructed based on the target network element's actual data sequence and predicted data sequence in the second historical period.

[0112] As mentioned above, the actual data sequence and the predicted data sequence of the target network element in the second historical period include the same number of network element evaluation index data. Therefore, the actual data sequence and the predicted data sequence of the target network element in the second historical period are equivalent to two one-dimensional series of the same length. Therefore, the actual data sequence and the predicted data sequence of the target network element in the second historical period can be subjected to series difference processing to obtain a historical error sequence. The so-called series difference refers to performing subtraction operations on the elements in the corresponding positions of two series of the same length to obtain a new series (i.e., a difference series). Therefore, the element value of the i-th element in the above-mentioned historical error sequence is the difference between the element value of the i-th element in the actual data sequence of the target network element in the second historical period and the element value of the i-th element in the predicted data sequence of the target network element in the second historical period.

[0113] S303: Perform statistical analysis on the historical error sequence to obtain historical local features of the historical error sequence.

[0114] Among them, the historical local characteristics include the historical local error mean and the historical local error standard deviation.

[0115] After obtaining the above-mentioned historical error sequence, the mean of the element values ​​of each element in the historical error sequence can be determined to obtain the historical local error mean, and the standard deviation of the element values ​​of each element in the historical error sequence can be determined to obtain the historical local error standard deviation; and then the above-mentioned historical local error mean and historical local error standard deviation can be used as the historical local characteristics of the above-mentioned historical error sequence.

[0116] S304: Determine a basic threshold based on historical local characteristics.

[0117] After obtaining the historical local features of the above historical error sequence, the basic threshold can be determined based on the historical local features.

[0118] Optionally, the sum of the above-mentioned historical local error mean and the historical local error standard deviation can be determined as the basic threshold; or, weights can be set for the above-mentioned historical local error mean and the historical local error standard deviation, and based on the set weights, the weighted sum of the above-mentioned historical local error mean and the historical local error standard deviation can be determined as the basic threshold.

[0119] In an optional embodiment, the basic threshold may be determined by the following formula:

[0120] td(t)=μ hist +k· hist

[0121] Among them, td(t) is the basic threshold, μ hist is the historical local error mean, σ hist is the historical local error standard deviation, and k is the confidence coefficient of the historical local error standard deviation. k can be set based on empirical values, test values ​​from multiple experiments, and actual application requirements. There is no specific limit on this, for example, k∈[2.5,3.5].

[0122] In this embodiment, a historical error sequence is constructed based on the actual data sequence and predicted data sequence of the target network element in the second historical period, and the historical local error mean and historical local error standard deviation are obtained by statistical analysis of the historical error sequence to determine the basic threshold. This can improve the matching of the physical meaning of the basic threshold and the current global error, thereby improving the credibility and reliability of the basic threshold, and further improving the accuracy of the network element fault detection finally obtained.

[0123] On the basis of the above embodiments, in an exemplary embodiment, the determination of the first correction value is further limited, optionally, as follows: Figure 4 As shown, the following steps may be included:

[0124] S401 : determining an error weighted value of a target network element according to a historical data sequence of the target network element in a global historical period and network element evaluation index data at a current moment.

[0125] In this embodiment, network element weights are set for the target network element and its associated network elements, wherein the network element weights can be set based on empirical values, test values ​​from multiple tests, and actual application requirements, and no specific limitation is made to this.

[0126] Optionally, determine the mean of each network element evaluation index data in the historical data sequence of the target network element in the global historical period as the first mean; determine the difference between the above first mean and the network element evaluation index data of the target network element at the current moment as the first difference; determine the product of the above first difference and the network element weight of the target network element as the error weighted value of the target network element.

[0127] Optionally, determine the mean of each network element evaluation index data of the target network element in the historical data sequence of the global historical period as the first mean, and determine the standard deviation of each network element evaluation index data of the target network element in the historical data sequence of the global historical period as the first standard deviation; determine the difference between the above-mentioned first mean and the network element evaluation index data of the target network element at the current moment as the first difference; determine the ratio of the above-mentioned first difference to the above-mentioned first standard deviation as the first ratio; determine the product of the above-mentioned first ratio and the network element weight of the target network element as the error weighted value of the target network element.

[0128] S402 : Determine an error weight value of the associated network element according to a historical data sequence of the associated network element of the target network element in a global historical period and network element evaluation index data at a current moment.

[0129] Optionally, determine the mean of each network element evaluation index data of the target network element's associated network elements in the historical data sequence of the global historical period as the second mean; determine the difference between the above second mean and the network element evaluation index data of the target network element's associated network elements at the current moment as the second difference; determine the product of the above second difference and the network element weight of the target network element's associated network elements as the error weighted value of the target network element's associated network elements.

[0130] Optionally, determine the mean of the evaluation index data of each network element of the target network element in the historical data sequence of the associated network elements in the global historical period as the second mean, and determine the standard deviation of the evaluation index data of each network element of the associated network elements in the historical data sequence of the target network element in the global historical period as the second standard deviation; determine the difference between the above-mentioned second mean and the network element evaluation index data of the associated network elements of the target network element at the current moment as the second difference; determine the ratio of the above-mentioned second difference to the above-mentioned second standard deviation as the second ratio; determine the product of the above-mentioned second ratio and the network element weight of the associated network elements of the target network element as the error weighted value of the associated network elements of the target network element.

[0131] S403: Determine a first correction value according to the average of the weighted error value of the target network element and the weighted error value of the associated network element.

[0132] After obtaining the above-mentioned error weighted value of the target network element and the error weighted value of the target network element's associated network elements, the average of the error weighted value of the target network element and the error weighted value of the target network element's associated network elements can be determined, and the obtained average is determined as the first correction value.

[0133] In this embodiment, the first correction value is determined based on the average of the error weighted value of the above-mentioned target network element and the error weighted value of the associated network element. When the above-mentioned basic threshold is adjusted according to the association of multiple network elements, the influence weight between the target network element and the associated network element is introduced, so that the obtained first correction value is more in line with the association relationship between the target network element and the associated network element in actual application, thereby improving the accuracy of the first correction value, improving the credibility and reliability of the obtained dynamic threshold, and thus improving the accuracy of the network element fault detection finally obtained.

[0134] On the basis of the above embodiments, in an exemplary embodiment, the determination of the error weighted value of the target network element is further refined, optionally, as follows: Figure 5 As shown, the following steps may be included:

[0135] S501 , performing statistical analysis on a historical data sequence of a target network element in a global historical period to obtain a historical global feature of the target network element.

[0136] Among them, the historical global characteristics include the historical global error mean and the historical global error standard deviation.

[0137] Optionally, first, based on the target network element's historical data sequence in the global historical period, a data sequence for the target network element in a future period of the global historical period is predicted as a first data sequence; wherein, the prediction method for the above-mentioned first data sequence is the same as the prediction method for the above-mentioned future data sequence and / or the predicted data sequence for the target network element in the second historical period, and will not be repeated here. Subsequently, since the target network element's historical data sequence in the global historical period and the above-mentioned first data sequence include the same number of network element evaluation indicator data, a sequence difference processing can be performed on the target network element's historical data sequence in the global historical period and the above-mentioned first data sequence to obtain a new sequence as a first error sequence. Then, the mean of the element values ​​of each element in the first error sequence can be determined to obtain the historical global error mean of the target network element, and the standard deviation of the element values ​​of each element in the first error sequence can be determined to obtain the historical global error standard deviation of the target network element. In this way, the historical global error mean and historical global error standard deviation of the target network element can be used as the historical global features of the target network element.

[0138] S502 : Determine the real-time error of the target network element at the current moment according to the difference between the network element evaluation index data of the target network element at the current moment and the network element evaluation index data at the future moment corresponding to the current moment.

[0139] When the future data sequence of the target network element in the future time period is predicted based on the current data sequence of the target network element in the current time period, there is a future time corresponding to the current time in the future time period. Wherein, the current time is the last time in the current time period, and the future time corresponding to the current time is the last time in the future time period. Then, the network element evaluation index data of the target network element at the future time corresponding to the current time can be obtained from the predicted future data sequence, that is, the last network element evaluation index data in the future data sequence can be obtained. Thus, the difference between the last time network element evaluation index data in the current data sequence (that is, the network element evaluation index data of the target network element at the current time) and the last time network element evaluation index data in the future data sequence (that is, the network element evaluation index data of the target network element at the future time corresponding to the current time) can be determined to obtain the real-time error of the target network element at the current time.

[0140] S503 : Determine an error weighted value of the target network element according to the historical global characteristics, real-time error, and network element weight of the target network element.

[0141] After obtaining the historical global characteristics of the target network element and the real-time error at the current moment, the network element weight of the target network element can be further determined, thereby determining the error weighted value of the target network element based on the historical global characteristics, real-time error and network element weight of the target network element.

[0142] Optionally, determine the difference between the real-time error of the target network element at the current moment and the historical global error mean of the target network element; determine the ratio of the above difference to the historical global error standard deviation of the target network element; determine the product of the above ratio and the network element weight of the target network element to obtain the error weighted value of the target network element.

[0143] In this embodiment, by determining the error weighted value of the target network element based on the historical global characteristics, real-time error and network element weight of the target network element, the matching of the error weighted value of the target network element participating in the first correction value and the physical meaning of the current global error can be improved, thereby improving the matching of the physical meaning of the first correction value and the current global error, so as to improve the matching of the dynamic threshold and the physical meaning of the current global error, thereby improving the credibility and reliability of the dynamic threshold and improving the accuracy of the network element fault detection finally obtained.

[0144] On the basis of the above embodiments, in an exemplary embodiment, the determination of the error weight value of the associated network element is further refined, optionally, as follows: Figure 6 As shown, the following steps may be included:

[0145] S601 , performing statistical analysis on historical data sequences of associated network elements in a global historical period to obtain historical global features of the associated network elements.

[0146] Among them, the historical global characteristics include the historical global error mean and the historical global error standard deviation.

[0147] Optionally, first, based on the historical data sequence of the associated network element in the global historical period, a data sequence of the associated network element in the future period of the global historical period is predicted as a second data sequence; wherein, the prediction method of the above-mentioned second data sequence is the same as the prediction method of the above-mentioned future data sequence and / or the predicted data sequence of the target network element in the second historical period, and is not further described here. Subsequently, since the historical data sequence of the associated network element in the global historical period and the above-mentioned second data sequence include the same number of network element evaluation indicator data, a sequence difference processing can be performed on the historical data sequence of the associated network element in the global historical period and the above-mentioned second data sequence to obtain a new sequence as a second error sequence. Then, the mean of the element values ​​of each element in the second error sequence can be determined to obtain the historical global error mean of the associated network element, and the standard deviation of the element values ​​of each element in the second error sequence can be determined to obtain the historical global error standard deviation of the associated network element. In this way, the historical global error mean and historical global error standard deviation of the associated network element can be used as the historical global features of the associated network element.

[0148] S602 : Determine the real-time error of the associated network element at the current moment according to the difference between the network element evaluation index data of the associated network element at the current moment and the network element evaluation index data at the future moment corresponding to the current moment.

[0149] As mentioned above, similar to the target network element, the network element evaluation index data sequence of the associated network element in the future time period can be predicted based on the network element evaluation index data sequence of the associated network element in the current time period, thereby obtaining the network element evaluation index data of the associated network element at the current moment from the network element evaluation index data sequence of the associated network element in the current time period, and obtaining the network element evaluation index data of the associated network element at the future moment corresponding to the current moment from the network element evaluation index data sequence of the associated network element in the future time period, and determining the difference between the above-mentioned network element evaluation index data of the associated network element at the current moment and the network element evaluation index data of the future moment corresponding to the current moment, so as to obtain the real-time error of the associated network element at the current moment.

[0150] S603 : Determine an error weight value of the associated network element according to the historical global characteristics, real-time error, and network element weight of the associated network element.

[0151] After obtaining the historical global characteristics and real-time error of the above-mentioned associated network element at the current moment, the network element weight of the associated network element can be further determined, thereby determining the error weighted value of the associated network element based on the historical global characteristics, real-time error and network element weight of the associated network element.

[0152] Optionally, determine the difference between the real-time error of the associated network element at the current moment and the historical global error mean of the associated network element; determine the ratio of the above difference to the historical global error standard deviation of the associated network element; determine the product of the above ratio and the network element weight of the associated network element to obtain the error weighted value of the associated network element.

[0153] In this embodiment, by determining the error weighted value of the associated network element based on the historical global characteristics, real-time error and network element weight of the associated network element, the matching of the error weighted value of the associated network element participating in the first correction value and the physical meaning of the current global error can be improved, thereby improving the matching of the physical meaning of the first correction value and the current global error, so as to improve the matching of the dynamic threshold and the physical meaning of the current global error, thereby improving the credibility and reliability of the dynamic threshold and improving the accuracy of the network element fault detection finally obtained.

[0154] In an optional embodiment, in the above Figure 5 and Figure 6 Based on the embodiment shown, the first correction value can be determined by the following formula:

[0155]

[0156] in,

[0157] c1 is the first correction value, N is the total number of the target network element and the associated network elements of the target network element; ω i is the network element weight of the target network element and the i-th network element in the associated network elements of the target network element; r i is the real-time error between the target network element and the i-th network element in the associated network element at the current moment; u i is the historical global error mean between the target network element and the i-th network element in the associated network elements of the target network element; σ i is the historical global error standard deviation of the target network element and the i-th network element among the target network element's associated network elements.

[0158] On the basis of the above embodiments, in an exemplary embodiment, the determination of the dynamic threshold is further limited, optionally, as follows: Figure 7 As shown, the following steps may be included:

[0159] S701 : Determine a current error standard deviation according to a first sub-data sequence and a second sub-data sequence corresponding to the first sub-data sequence in a future data sequence.

[0160] The first sub-data sequence is a data sequence in the current data sequence that includes network element evaluation index data at the current moment.

[0161] To improve the accuracy of the dynamic threshold, the dynamic fluctuation characteristics of the target network element's network element evaluation index data can be incorporated into the revision of the basic threshold. It is understandable that due to the continuity of time, the target network element's network element evaluation index data is also interrelated and influences each other in time series. Therefore, to improve the accuracy of the revised value representing the dynamic fluctuation characteristics of the network element evaluation index data, it can be determined using network element evaluation index data over a shorter time period.

[0162] Based on this, a data sequence including network element evaluation index data at the current moment can be determined from the current data sequence as a first sub-data sequence, wherein the number of network element evaluation index data in the first sub-sequence is less than the number of network element evaluation index data in the current data sequence. Furthermore, a second sub-data sequence corresponding to the first sub-data sequence can be determined from the future data sequence, wherein the second sub-data sequence includes network element evaluation index data at a future moment corresponding to the current moment. The first sub-data sequence and the second sub-data sequence include network element evaluation index data at at least two consecutive moments, and the first sub-data sequence and the second sub-data sequence include the same number of network element evaluation index data.

[0163] For example, the current data sequence includes 50 network element evaluation index data, and the first sub-data sequence includes 10 network element evaluation index data from the 41st to the 50th in the current data sequence; correspondingly, the future data sequence also includes 50 network element evaluation index data, and the second sub-data sequence includes 10 network element evaluation index data from the 41st to the 50th in the future data sequence.

[0164] Then, the current error standard deviation is determined based on the first sub-data sequence and the second sub-data sequence.

[0165] Optionally, a difference process is performed on the first sub-data sequence and the second sub-data sequence to obtain an error data sequence, and the standard deviation of the element value of each element in the error sequence is determined to obtain the current error standard deviation.

[0166] Optionally, as described above, the element value of the i-th element in the current error sequence is the difference between the element value of the i-th element in the current data sequence and the element value of the i-th element in the future data sequence. Therefore, when constructing the current error sequence based on the current data sequence and the future data sequence, a current error subsequence corresponding to the first sub-data sequence can be obtained from the current error sequence, and the standard deviation of the element values ​​of each element in the current error subsequence can be determined to obtain the current error standard deviation. The current error subsequence is a subsequence in the current error sequence that includes the last element and has the same length as the first sub-data sequence. For example, the current data sequence includes 50 network element evaluation indicator data, and the first sub-data sequence includes 10 network element evaluation indicator data items (numbers 41-50) in the current data sequence. Accordingly, the second sub-data sequence includes 10 network element evaluation indicator data items (numbers 41-50) in the future data sequence, and the current error subsequence includes 10 elements (numbers 41-50) in the current error sequence.

[0167] S702: Determine a second correction value according to the current error standard deviation and the historical local error standard deviation.

[0168] As previously described, in determining the historical local error standard deviation, the target network element's predicted data sequence for the second historical period is first predicted based on its historical data sequence for the first historical period. A historical error sequence is then constructed based on the target network element's actual data sequence and predicted data sequence for the second historical period. The standard deviation of each element in the historical error sequence is then determined to obtain the historical local error standard deviation. Thus, after obtaining the current error standard deviation, the second correction value can be determined based on the current error standard deviation and the historical local error standard deviation.

[0169] Optionally, the ratio of the current error standard deviation to the historical local error standard deviation is determined to obtain the second correction value.

[0170] In an optional embodiment, the second correction value can be determined by the following formula:

[0171]

[0172] Among them, c2 is the second correction value, σ win (t) is the current error standard deviation, σ hist is the historical local error standard deviation, and α is the fluctuation sensitivity coefficient. α can be set based on empirical values, test values ​​from multiple experiments, and actual application requirements. There is no specific limit on this, for example, α∈[0.5,1.5].

[0173] S703: Determine a third correction value according to whether the current time period is within a designated event period and the designated event coefficient.

[0174] The designated event coefficient is determined based on a historical data sequence of the target network element within the designated event period; the designated events include network element cutover events and / or non-working day events.

[0175] The so-called network element cutover refers to the process of migrating services from old network elements to new network elements through technical means.

[0176] Non-working days are days that fall outside of "working hours" under specific rules or regulations. Their core function is to distinguish between time required for regular work or study and time available for discretionary use or rest. Essentially, they serve as a way to divide time resources. Non-working days may also include statutory holidays and public holidays.

[0177] After determining the current time period, it is possible to determine whether the current time period falls within a specified event period, that is, to determine whether the current time period falls within a specified time period. For example, the current time period can be determined to fall within a network element cutover event period, where the so-called network element cutover event period refers to the implementation period within the full process cycle of a network element cutover. Another example is to determine whether the current time period falls within a non-working day event period, that is, to determine whether the natural day in which the current time period falls is a non-working day.

[0178] Optionally, if the specified events only include network element cutover events, it is only necessary to determine whether the current time period is in the network element cutover event period; or, if the specified events only include non-working day events, it is only necessary to determine whether the current time period is in the non-working day event period; or, if the specified events include network element cutover events and non-working day events, it is necessary to determine whether the current time period is in the network element cutover event period and whether it is in the non-working day event period.

[0179] Furthermore, it is understood that the target network element may be in a designated event period in the historical period before the current period. Thus, network element evaluation index data for each consecutive moment in the target network element's history within the designated event period can be obtained to obtain a historical data sequence of the target network element's history within the designated event period. Furthermore, the designated event coefficient can be determined based on this historical data sequence.

[0180] Optionally, obtain a historical data sequence of the target network element when its history is in a specified event period, and based on the above historical data sequence, predict a historical data sequence of the target network element in a future period when its history is in the specified event period, and then perform sequence difference processing on the historical data sequence of the target network element when its history is in the specified event period and the historical data sequence of the future period when its history is in the specified event period to obtain a specified error sequence of the target network element, and use a Lasso regression model (Lasso Regression, a linear regression model with L1 regularization) to perform significant variable screening and coefficient fitting on the above specified error sequence to obtain the specified event coefficient.

[0181] Then, according to whether the current time period is in the designated event period and the designated event coefficient, a third correction value is determined.

[0182] Optionally, the product of a parameter value of a parameter used to characterize the situation that the current time period is in a specified event period and a specified event coefficient is determined to obtain a third correction value.

[0183] In an optional embodiment, the designated event includes a network element cutover event and a non-working day event, and the third correction value may be determined by the following formula.

[0184] c3=1+β1·day t +β2·Event t

[0185] Among them, c3 is the third correction value; day t It is used to indicate that the current period is not a holiday event period. If the current period is not a working day event period, then day t =1, if the current period is not in the non-working day event period (i.e. in the working day event period), then day t =0; β1 is the specified event coefficient determined based on the historical data sequence of the target network element in the non-working day event period; Event t Used to indicate that the current time period is in the network element cutover event period. If the current time period is in the network element cutover event period, then Event t =1, if the current time period is in the NE cutover event period, then Event t =0; β2 is a designated event coefficient determined according to a historical data sequence of the target network element being within the network element cutover event period.

[0186] S704: Use the first correction value and the auxiliary value to adjust the basic threshold to obtain a dynamic threshold.

[0187] The auxiliary value includes the second correction value and / or the third correction value.

[0188] After obtaining the above-mentioned second correction value and third correction value, the second correction value can be determined as an auxiliary value, the third correction value can be determined as an auxiliary value, or both the second correction value and the third correction value can be determined as correction values, and then the first correction value and the auxiliary value are used to adjust the basic threshold to obtain a dynamic threshold.

[0189] Optionally, the sum of the first correction value and the auxiliary value may be determined to obtain a dynamic threshold.

[0190] In an optional embodiment, the dynamic threshold may be determined by the following formula:

[0191]

[0192] in,

[0193] Td(t) is the basic threshold, μ hist is the historical local error mean, σ hist is the historical local error standard deviation, k is the confidence coefficient of the historical local error standard deviation; N is the total number of target network elements and the associated network elements of the target network element; ω i is the network element weight of the target network element and the i-th network element in the associated network elements of the target network element; r i is the real-time error between the target network element and the i-th network element in the associated network element at the current moment; u i is the historical global error mean between the target network element and the i-th network element in the associated network elements of the target network element; σ i is the historical global error standard deviation of the target network element and the i-th network element in the associated network elements of the target network element; σ win (t) is the current error standard deviation, α is the fluctuation sensitivity coefficient; day t It is used to indicate that the current period is not a holiday event period. If the current period is not a working day event period, then day t =1, if the current period is not a non-working day event period, then day t =0; β1 is the specified event coefficient determined based on the historical data sequence of the target network element in the non-working day event period; Event t Used to indicate that the current time period is in the network element cutover event period. If the current time period is in the network element cutover event period, then Event t=1, if the current time period is in the NE cutover event period, then Event t =0; β2 is a designated event coefficient determined according to a historical data sequence of the target network element being within the network element cutover event period.

[0194] In this embodiment, by determining the second correction value and the third correction value, the dynamic fluctuation characteristics of the network element evaluation index data of the target network element and the environmental perception characteristics of the target network element based on the specified event period are introduced to adjust the basic threshold. Therefore, the accuracy of the determined dynamic threshold can be improved by real-time perception of the network status, service characteristics and historical laws, thereby providing more comprehensive and accurate detection of network element faults and improving the accuracy of network element fault detection.

[0195] Based on the above embodiments, in an exemplary embodiment, Figure 8 As shown, the following steps may be included:

[0196] S801 , predicting a future data sequence of the target network element in a future time period based on the current data sequence of the target network element in the current time period, and determining a current global error between the current data sequence and the future data sequence.

[0197] S802, based on the data generation model, predict the predicted data sequence of the target network element in the second historical period according to the historical data sequence of the target network element in the first historical period, and construct a historical error sequence according to the actual data sequence and the predicted data sequence of the target network element in the second historical period.

[0198] S803: Perform statistical analysis on the historical error sequence to obtain historical local features of the historical error sequence, and determine a basic threshold based on the historical local features.

[0199] S804: Perform statistical analysis on the historical data sequence of the target network element in the global historical period to obtain the historical global characteristics of the target network element.

[0200] S805 , determining the real-time error of the target network element at the current moment according to the difference between the network element evaluation index data of the target network element at the current moment and the network element evaluation index data at the future moment corresponding to the current moment.

[0201] S806 , determining an error weighted value of the target network element according to the historical global characteristics, real-time error, and network element weight of the target network element.

[0202] S807 , performing statistical analysis on the historical data series of the associated network elements in the global historical period to obtain historical global features of the associated network elements.

[0203] S808 , determining a real-time error of the associated network element at the current moment according to a difference between the network element evaluation index data of the associated network element at the current moment and the network element evaluation index data at a future moment corresponding to the current moment.

[0204] S809 , determining an error weighted value of the associated network element according to the historical global characteristics, real-time error, and network element weight of the associated network element.

[0205] S810 : Determine a first correction value according to a mean value between the weighted error value of the target network element and the weighted error value of the associated network element.

[0206] S811, determine a current error standard deviation based on the first sub-data sequence and a second sub-data sequence corresponding to the first sub-data sequence in the future data sequence, and determine a second correction value based on the current error standard deviation and a historical local error standard deviation.

[0207] S812: Determine a third correction value according to whether the current time period is within a designated event period and the designated event coefficient.

[0208] S813: Use the first correction value and the auxiliary value to adjust the basic threshold to obtain a dynamic threshold.

[0209] The specific implementation of S801-S813 is the same as that in the above method embodiments, and will not be repeated here.

[0210] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0211] Based on the same inventive concept, embodiments of the present application further provide a network element fault detection device for implementing the aforementioned network element fault detection method. The implementation solution provided by this device is similar to the implementation solution described in the aforementioned method. Therefore, the specific limitations in one or more embodiments of the network element fault detection device provided below can be found in the above-mentioned limitations on the network element fault detection method and will not be further elaborated here.

[0212] In an exemplary embodiment, Figure 9 As shown, a network element fault detection device is provided, including: a data prediction module 910, an error determination module 920, a threshold determination module 930, a correction value determination module 940, a threshold adjustment module 950 and a fault detection module 960, wherein:

[0213] The data prediction module 910 is configured to predict a future data sequence of the target network element in a future time period based on the current data sequence of the target network element in the current time period; wherein the data sequence includes network element evaluation index data of at least two consecutive moments;

[0214] an error determination module 920 for determining a current global error between a current data sequence and a future data sequence;

[0215] A threshold determination module 930 is configured to process a historical data sequence of the target network element in a first historical period to obtain a basic threshold; wherein the first historical period is a period when the target network element is in a normal state;

[0216] Correction value determination module 940 is configured to determine a first correction value based on a historical data sequence of the target network element in a global historical period and network element evaluation index data at a current moment, and a historical data sequence of network elements associated with the target network element in a global historical period and network element evaluation index data at a current moment; wherein the global historical period is a historical period corresponding to the current period, and the global historical period includes the first historical period;

[0217] A threshold adjustment module 950 is configured to adjust the basic threshold using the first correction value to obtain a dynamic threshold;

[0218] The fault detection module 960 is configured to determine a fault detection result of a target network element according to a magnitude relationship between a current global error and a dynamic threshold.

[0219] In an exemplary embodiment, the threshold determination module 930 is specifically configured to:

[0220] Based on the data generation model, the predicted data sequence of the target network element in the second historical period is predicted according to the historical data sequence of the target network element in the first historical period; the historical error sequence is constructed according to the actual data sequence and the predicted data sequence of the target network element in the second historical period; the historical error sequence is statistically analyzed to obtain the historical local characteristics of the historical error sequence; wherein the historical local characteristics include the historical local error mean and the historical local error standard deviation; and the basic threshold is determined according to the historical local characteristics.

[0221] In an exemplary embodiment, the correction value determination module 940 includes:

[0222] A first determination submodule is configured to determine an error weighted value of a target network element based on a historical data sequence of the target network element in a global historical period and network element evaluation index data at a current moment;

[0223] A second determining submodule is configured to determine an error weight value of an associated network element according to a historical data sequence of an associated network element of the target network element in a global historical period and network element evaluation index data at a current moment;

[0224] The third determining submodule is configured to determine a first correction value according to a mean value between the weighted error value of the target network element and the weighted error value of the associated network element.

[0225] In an exemplary embodiment, the first determining submodule is specifically configured to:

[0226] Perform statistical analysis on the historical data series of the target network element in the global historical period to obtain the historical global characteristics of the target network element; wherein the historical global characteristics include the historical global error mean and the historical global error standard deviation; determine the real-time error of the target network element at the current moment based on the difference between the network element evaluation index data of the target network element at the current moment and the network element evaluation index data at the future moment corresponding to the current moment; determine the error weighted value of the target network element based on the historical global characteristics, real-time error and network element weight of the target network element.

[0227] In an exemplary embodiment, the second determining submodule is specifically configured to:

[0228] Statistical analysis is performed on the historical data series of the associated network elements in the global historical period to obtain the historical global characteristics of the associated network elements; wherein the historical global characteristics include the historical global error mean and the historical global error standard deviation; based on the difference between the network element evaluation index data of the associated network element at the current moment and the network element evaluation index data at the future moment corresponding to the current moment, the real-time error of the associated network element at the current moment is determined; based on the historical global characteristics, real-time error and network element weight of the associated network element, the error weighted value of the associated network element is determined.

[0229] In an exemplary embodiment, the threshold adjustment module 950 is specifically configured to:

[0230] Determine the current error standard deviation based on the first sub-data sequence and the second sub-data sequence corresponding to the first sub-data sequence in the future data sequence; wherein the first sub-data sequence is a data sequence in the current data sequence that includes the network element evaluation index data at the current moment; determine the second correction value based on the current error standard deviation and the historical local error standard deviation; determine the third correction value based on the situation that the current time period is in a specified event period, and the specified event coefficient; wherein the specified event coefficient is determined based on the historical data sequence of the target network element in the specified event period; the specified event includes a network element cutover event and / or a non-working day event; use the first correction value and the auxiliary value to adjust the basic threshold to obtain a dynamic threshold; wherein the auxiliary value includes the second correction value and / or the third correction value.

[0231] Each module in the above-mentioned network element fault detection device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in the computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0232] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Figure 10 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store network element evaluation index data of network elements. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a network element fault detection method is implemented.

[0233] Those skilled in the art will understand that Figure 10 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0234] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in each method embodiment of the above-mentioned network element detection method when executing the computer program.

[0235] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in each method embodiment of the above-mentioned network element detection method are implemented.

[0236] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in each method embodiment of the above-mentioned network element detection method when executed by a processor.

[0237] It should be noted that the data involved in this application (including but not limited to the network element evaluation indicators of the target network element and the associated network elements of the target network element, etc.) are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0238] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processor involved in the various embodiments provided herein may be, but are not limited to, a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, and the like.

[0239] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0240] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A network element fault detection method, characterized in that: The method comprises: Predicting a future data sequence of the target network element in a future time period based on the current data sequence of the target network element in the current time period; wherein the data sequence includes network element evaluation index data of at least two consecutive moments; determining a current global error between the current data sequence and the future data sequence; Processing a historical data sequence of the target network element in a first historical period to obtain a basic threshold value; wherein the first historical period is a period during which the target network element is in a normal state; Determining a first correction value based on a historical data sequence of the target network element in a global historical period and network element evaluation index data at a current moment, and a historical data sequence of network elements associated with the target network element in a global historical period and network element evaluation index data at a current moment; wherein the global historical period is a historical period corresponding to the current period, and the global historical period includes the first historical period; Using the first correction value to adjust the basic threshold to obtain a dynamic threshold; A fault detection result of the target network element is determined according to a magnitude relationship between the current global error and the dynamic threshold.

2. The method according to claim 1, characterized in that The processing of the historical data sequence of the target network element in the first historical period to obtain the basic threshold value includes: Based on the data generation model, predicting the predicted data sequence of the target network element in the second historical period according to the historical data sequence of the target network element in the first historical period; constructing a historical error sequence based on the actual data sequence and the predicted data sequence of the target network element in the second historical period; Performing statistical analysis on the historical error sequence to obtain historical local features of the historical error sequence; wherein the historical local features include a historical local error mean and a historical local error standard deviation; A basic threshold is determined according to the historical local characteristics.

3. The method according to claim 1, characterized in that Determining the first correction value based on the historical data sequence of the target network element in the global historical period and the network element evaluation index data at the current moment, and the historical data sequence of the network elements associated with the target network element in the global historical period and the network element evaluation index data at the current moment, includes: Determining an error weighted value of the target network element based on a historical data sequence of the target network element in a global historical period and network element evaluation index data at a current moment; Determining an error weighted value of the associated network element according to a historical data sequence of the associated network element of the target network element in a global historical period and the network element evaluation index data at the current moment; A first correction value is determined according to a mean value between the weighted error value of the target network element and the weighted error value of the associated network element.

4. The method according to claim 3, characterized in that The determining the error weighted value of the target network element according to the historical data sequence of the target network element in the global historical period and the network element evaluation index data at the current moment includes: Performing statistical analysis on a historical data sequence of the target network element in a global historical period to obtain a historical global feature of the target network element; wherein the historical global feature includes a historical global error mean and a historical global error standard deviation; Determining a real-time error of the target network element at the current moment according to a difference between the network element evaluation index data of the target network element at the current moment and the network element evaluation index data at a future moment corresponding to the current moment; An error weighted value of the target network element is determined according to the historical global characteristics, the real-time error and the network element weight of the target network element.

5. The method according to claim 3, characterized in that The determining, based on a historical data sequence of an associated network element of the target network element in a global historical period and the network element evaluation index data at the current moment, an error weighted value of the associated network element includes: Performing statistical analysis on the historical data series of the associated network element in the global historical period to obtain historical global characteristics of the associated network element; wherein the historical global characteristics include a historical global error mean and a historical global error standard deviation; Determining a real-time error of the associated network element at the current moment according to a difference between the network element evaluation index data of the associated network element at the current moment and the network element evaluation index data at a future moment corresponding to the current moment; An error weighted value of the associated network element is determined according to the historical global characteristics, real-time error and network element weight of the associated network element.

6. The method according to claim 2, characterized in that The step of adjusting the basic threshold using the first correction value to obtain a dynamic threshold includes: Determining a current error standard deviation based on a first sub-data sequence and a second sub-data sequence corresponding to the first sub-data sequence in the future data sequence; wherein the first sub-data sequence is a data sequence in the current data sequence that includes network element evaluation indicator data at a current moment; determining a second correction value according to the current error standard deviation and the historical local error standard deviation; Determining a third correction value based on the current time period being within a designated event period and a designated event coefficient; wherein the designated event coefficient is determined based on a historical data sequence of the target network element being within the designated event period; the designated event includes a network element cutover event and / or a non-working day event; The basic threshold is adjusted using the first correction value and the auxiliary value to obtain a dynamic threshold; wherein the auxiliary value includes the second correction value and / or the third correction value.

7. A network element fault detection device, characterized in that: The device comprises: A data prediction module, configured to predict a future data sequence of the target network element in a future time period based on the current data sequence of the target network element in the current time period; wherein the data sequence includes network element evaluation index data of at least two consecutive moments; an error determination module, configured to determine a current global error between the current data sequence and the future data sequence; a threshold determination module, configured to process a historical data sequence of the target network element in a first historical period to obtain a basic threshold; wherein the first historical period is a period during which the target network element is in a normal state; a correction value determination module, configured to determine a first correction value based on a historical data sequence of the target network element in a global historical period and network element evaluation index data at a current moment, and a historical data sequence of network elements associated with the target network element in a global historical period and network element evaluation index data at a current moment; wherein the global historical period is a historical period corresponding to the current period, and the global historical period includes the first historical period; a threshold adjustment module, configured to adjust the basic threshold using the first correction value to obtain a dynamic threshold; The fault detection module is used to determine the fault detection result of the target network element according to the size relationship between the current global error and the dynamic threshold.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images